[mirror_ubuntu-artful-kernel.git] / kernel / sched_stats.h


#ifdef CONFIG_SCHEDSTATS
/*
 * bump this up when changing the output format or the meaning of an existing
 * format, so that tools can adapt (or abort)
 */
#define SCHEDSTAT_VERSION 14

static int show_schedstat(struct seq_file *seq, void *v)
{
	int cpu;

	seq_printf(seq, "version %d\n", SCHEDSTAT_VERSION);
	seq_printf(seq, "timestamp %lu\n", jiffies);
	for_each_online_cpu(cpu) {
		struct rq *rq = cpu_rq(cpu);
#ifdef CONFIG_SMP
		struct sched_domain *sd;
		int dcount = 0;
#endif

		/* runqueue-specific stats */
		seq_printf(seq,
		    "cpu%d %u %u %u %u %u %u %u %u %u %llu %llu %lu",
		    cpu, rq->yld_both_empty,
		    rq->yld_act_empty, rq->yld_exp_empty, rq->yld_count,
		    rq->sched_switch, rq->sched_count, rq->sched_goidle,
		    rq->ttwu_count, rq->ttwu_local,
		    rq->rq_sched_info.cpu_time,
		    rq->rq_sched_info.run_delay, rq->rq_sched_info.pcount);

		seq_printf(seq, "\n");

#ifdef CONFIG_SMP
		/* domain-specific stats */
		preempt_disable();
		for_each_domain(cpu, sd) {
			enum cpu_idle_type itype;
			char mask_str[NR_CPUS];

			cpumask_scnprintf(mask_str, NR_CPUS, sd->span);
			seq_printf(seq, "domain%d %s", dcount++, mask_str);
			for (itype = CPU_IDLE; itype < CPU_MAX_IDLE_TYPES;
					itype++) {
				seq_printf(seq, " %u %u %u %u %u %u %u %u",
				    sd->lb_count[itype],
				    sd->lb_balanced[itype],
				    sd->lb_failed[itype],
				    sd->lb_imbalance[itype],
				    sd->lb_gained[itype],
				    sd->lb_hot_gained[itype],
				    sd->lb_nobusyq[itype],
				    sd->lb_nobusyg[itype]);
			}
			seq_printf(seq, " %u %u %u %u %u %u %u %u %u %u %u %u\n",
			    sd->alb_count, sd->alb_failed, sd->alb_pushed,
			    sd->sbe_count, sd->sbe_balanced, sd->sbe_pushed,
			    sd->sbf_count, sd->sbf_balanced, sd->sbf_pushed,
			    sd->ttwu_wake_remote, sd->ttwu_move_affine,
			    sd->ttwu_move_balance);
		}
		preempt_enable();
#endif
	}
	return 0;
}

static int schedstat_open(struct inode *inode, struct file *file)
{
	unsigned int size = PAGE_SIZE * (1 + num_online_cpus() / 32);
	char *buf = kmalloc(size, GFP_KERNEL);
	struct seq_file *m;
	int res;

	if (!buf)
		return -ENOMEM;
	res = single_open(file, show_schedstat, NULL);
	if (!res) {
		m = file->private_data;
		m->buf = buf;
		m->size = size;
	} else
		kfree(buf);
	return res;
}

const struct file_operations proc_schedstat_operations = {
	.open    = schedstat_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = single_release,
};

/*
 * Expects runqueue lock to be held for atomicity of update
 */
static inline void
rq_sched_info_arrive(struct rq *rq, unsigned long long delta)
{
	if (rq) {
		rq->rq_sched_info.run_delay += delta;
		rq->rq_sched_info.pcount++;
	}
}

/*
 * Expects runqueue lock to be held for atomicity of update
 */
static inline void
rq_sched_info_depart(struct rq *rq, unsigned long long delta)
{
	if (rq)
		rq->rq_sched_info.cpu_time += delta;
}
# define schedstat_inc(rq, field)	do { (rq)->field++; } while (0)
# define schedstat_add(rq, field, amt)	do { (rq)->field += (amt); } while (0)
# define schedstat_set(var, val)	do { var = (val); } while (0)
#else /* !CONFIG_SCHEDSTATS */
static inline void
rq_sched_info_arrive(struct rq *rq, unsigned long long delta)
{}
static inline void
rq_sched_info_depart(struct rq *rq, unsigned long long delta)
{}
# define schedstat_inc(rq, field)	do { } while (0)
# define schedstat_add(rq, field, amt)	do { } while (0)
# define schedstat_set(var, val)	do { } while (0)
#endif

#ifdef CONFIG_SCHEDSTATS
/*
 * Called when a process is dequeued from the active array and given
 * the cpu.  We should note that with the exception of interactive
 * tasks, the expired queue will become the active queue after the active
 * queue is empty, without explicitly dequeuing and requeuing tasks in the
 * expired queue.  (Interactive tasks may be requeued directly to the
 * active queue, thus delaying tasks in the expired queue from running;
 * see scheduler_tick()).
 *
 * This function is only called from sched_info_arrive(), rather than
 * dequeue_task(). Even though a task may be queued and dequeued multiple
 * times as it is shuffled about, we're really interested in knowing how
 * long it was from the *first* time it was queued to the time that it
 * finally hit a cpu.
 */
static inline void sched_info_dequeued(struct task_struct *t)
{
	t->sched_info.last_queued = 0;
}

/*
 * Called when a task finally hits the cpu.  We can now calculate how
 * long it was waiting to run.  We also note when it began so that we
 * can keep stats on how long its timeslice is.
 */
static void sched_info_arrive(struct task_struct *t)
{
	unsigned long long now = sched_clock(), delta = 0;

	if (t->sched_info.last_queued)
		delta = now - t->sched_info.last_queued;
	sched_info_dequeued(t);
	t->sched_info.run_delay += delta;
	t->sched_info.last_arrival = now;
	t->sched_info.pcount++;

	rq_sched_info_arrive(task_rq(t), delta);
}

/*
 * Called when a process is queued into either the active or expired
 * array.  The time is noted and later used to determine how long we
 * had to wait for us to reach the cpu.  Since the expired queue will
 * become the active queue after active queue is empty, without dequeuing
 * and requeuing any tasks, we are interested in queuing to either. It
 * is unusual but not impossible for tasks to be dequeued and immediately
 * requeued in the same or another array: this can happen in sched_yield(),
 * set_user_nice(), and even load_balance() as it moves tasks from runqueue
 * to runqueue.
 *
 * This function is only called from enqueue_task(), but also only updates
 * the timestamp if it is already not set.  It's assumed that
 * sched_info_dequeued() will clear that stamp when appropriate.
 */
static inline void sched_info_queued(struct task_struct *t)
{
	if (unlikely(sched_info_on()))
		if (!t->sched_info.last_queued)
			t->sched_info.last_queued = sched_clock();
}

/*
 * Called when a process ceases being the active-running process, either
 * voluntarily or involuntarily.  Now we can calculate how long we ran.
 */
static inline void sched_info_depart(struct task_struct *t)
{
	unsigned long long delta = sched_clock() - t->sched_info.last_arrival;

	t->sched_info.cpu_time += delta;
	rq_sched_info_depart(task_rq(t), delta);
}

/*
 * Called when tasks are switched involuntarily due, typically, to expiring
 * their time slice.  (This may also be called when switching to or from
 * the idle task.)  We are only called when prev != next.
 */
static inline void
__sched_info_switch(struct task_struct *prev, struct task_struct *next)
{
	struct rq *rq = task_rq(prev);

	/*
	 * prev now departs the cpu.  It's not interesting to record
	 * stats about how efficient we were at scheduling the idle
	 * process, however.
	 */
	if (prev != rq->idle)
		sched_info_depart(prev);

	if (next != rq->idle)
		sched_info_arrive(next);
}
static inline void
sched_info_switch(struct task_struct *prev, struct task_struct *next)
{
	if (unlikely(sched_info_on()))
		__sched_info_switch(prev, next);
}
#else
#define sched_info_queued(t)		do { } while (0)
#define sched_info_switch(t, next)	do { } while (0)
#endif /* CONFIG_SCHEDSTATS */
Commit	Line	Data
425e0968 IM	1
	2	#ifdef CONFIG_SCHEDSTATS
	3	/*
	4	* bump this up when changing the output format or the meaning of an existing
	5	* format, so that tools can adapt (or abort)
	6	*/
	7	#define SCHEDSTAT_VERSION 14
	8
	9	static int show_schedstat(struct seq_file seq, void v)
	10	{
	11	int cpu;
	12
	13	seq_printf(seq, "version %d\n", SCHEDSTAT_VERSION);
	14	seq_printf(seq, "timestamp %lu\n", jiffies);
	15	for_each_online_cpu(cpu) {
	16	struct rq *rq = cpu_rq(cpu);
	17	#ifdef CONFIG_SMP
	18	struct sched_domain *sd;
2d72376b	19	int dcount = 0;
425e0968 IM	20	#endif
	21
	22	/* runqueue-specific stats */
	23	seq_printf(seq,
480b9434	24	"cpu%d %u %u %u %u %u %u %u %u %u %llu %llu %lu",
425e0968	25	cpu, rq->yld_both_empty,
2d72376b IM	26	rq->yld_act_empty, rq->yld_exp_empty, rq->yld_count,
	27	rq->sched_switch, rq->sched_count, rq->sched_goidle,
	28	rq->ttwu_count, rq->ttwu_local,
425e0968	29	rq->rq_sched_info.cpu_time,
2d72376b	30	rq->rq_sched_info.run_delay, rq->rq_sched_info.pcount);
425e0968 IM	31
	32	seq_printf(seq, "\n");
	33
	34	#ifdef CONFIG_SMP
	35	/* domain-specific stats */
	36	preempt_disable();
	37	for_each_domain(cpu, sd) {
	38	enum cpu_idle_type itype;
	39	char mask_str[NR_CPUS];
	40
	41	cpumask_scnprintf(mask_str, NR_CPUS, sd->span);
2d72376b	42	seq_printf(seq, "domain%d %s", dcount++, mask_str);
425e0968 IM	43	for (itype = CPU_IDLE; itype < CPU_MAX_IDLE_TYPES;
425e0968 IM	44	itype++) {
480b9434	45	seq_printf(seq, " %u %u %u %u %u %u %u %u",
2d72376b	46	sd->lb_count[itype],
425e0968 IM	47	sd->lb_balanced[itype],
	48	sd->lb_failed[itype],
	49	sd->lb_imbalance[itype],
	50	sd->lb_gained[itype],
	51	sd->lb_hot_gained[itype],
	52	sd->lb_nobusyq[itype],
	53	sd->lb_nobusyg[itype]);
	54	}
480b9434	55	seq_printf(seq, " %u %u %u %u %u %u %u %u %u %u %u %u\n",
2d72376b IM	56	sd->alb_count, sd->alb_failed, sd->alb_pushed,
	57	sd->sbe_count, sd->sbe_balanced, sd->sbe_pushed,
	58	sd->sbf_count, sd->sbf_balanced, sd->sbf_pushed,
425e0968 IM	59	sd->ttwu_wake_remote, sd->ttwu_move_affine,
	60	sd->ttwu_move_balance);
	61	}
	62	preempt_enable();
	63	#endif
	64	}
	65	return 0;
	66	}
	67
	68	static int schedstat_open(struct inode inode, struct file file)
	69	{
	70	unsigned int size = PAGE_SIZE * (1 + num_online_cpus() / 32);
	71	char *buf = kmalloc(size, GFP_KERNEL);
	72	struct seq_file *m;
	73	int res;
	74
	75	if (!buf)
	76	return -ENOMEM;
	77	res = single_open(file, show_schedstat, NULL);
	78	if (!res) {
	79	m = file->private_data;
	80	m->buf = buf;
	81	m->size = size;
	82	} else
	83	kfree(buf);
	84	return res;
	85	}
	86
	87	const struct file_operations proc_schedstat_operations = {
	88	.open = schedstat_open,
	89	.read = seq_read,
	90	.llseek = seq_lseek,
	91	.release = single_release,
	92	};
	93
	94	/*
	95	* Expects runqueue lock to be held for atomicity of update
	96	*/
	97	static inline void
	98	rq_sched_info_arrive(struct rq *rq, unsigned long long delta)
	99	{
	100	if (rq) {
	101	rq->rq_sched_info.run_delay += delta;
2d72376b	102	rq->rq_sched_info.pcount++;
425e0968 IM	103	}
	104	}
	105
	106	/*
	107	* Expects runqueue lock to be held for atomicity of update
	108	*/
	109	static inline void
	110	rq_sched_info_depart(struct rq *rq, unsigned long long delta)
	111	{
	112	if (rq)
	113	rq->rq_sched_info.cpu_time += delta;
	114	}
	115	# define schedstat_inc(rq, field) do { (rq)->field++; } while (0)
	116	# define schedstat_add(rq, field, amt) do { (rq)->field += (amt); } while (0)
c3c70119	117	# define schedstat_set(var, val) do { var = (val); } while (0)
425e0968 IM	118	#else /* !CONFIG_SCHEDSTATS */
	119	static inline void
	120	rq_sched_info_arrive(struct rq *rq, unsigned long long delta)
	121	{}
	122	static inline void
	123	rq_sched_info_depart(struct rq *rq, unsigned long long delta)
	124	{}
	125	# define schedstat_inc(rq, field) do { } while (0)
	126	# define schedstat_add(rq, field, amt) do { } while (0)
c3c70119	127	# define schedstat_set(var, val) do { } while (0)
425e0968 IM	128	#endif
425e0968 IM	129
75d4ef16	130	#ifdef CONFIG_SCHEDSTATS
425e0968 IM	131	/*
	132	* Called when a process is dequeued from the active array and given
	133	* the cpu. We should note that with the exception of interactive
	134	* tasks, the expired queue will become the active queue after the active
	135	* queue is empty, without explicitly dequeuing and requeuing tasks in the
	136	* expired queue. (Interactive tasks may be requeued directly to the
	137	* active queue, thus delaying tasks in the expired queue from running;
	138	* see scheduler_tick()).
	139	*
	140	* This function is only called from sched_info_arrive(), rather than
	141	* dequeue_task(). Even though a task may be queued and dequeued multiple
	142	* times as it is shuffled about, we're really interested in knowing how
	143	* long it was from the first time it was queued to the time that it
	144	* finally hit a cpu.
	145	*/
	146	static inline void sched_info_dequeued(struct task_struct *t)
	147	{
	148	t->sched_info.last_queued = 0;
	149	}
	150
	151	/*
	152	* Called when a task finally hits the cpu. We can now calculate how
	153	* long it was waiting to run. We also note when it began so that we
	154	* can keep stats on how long its timeslice is.
	155	*/
	156	static void sched_info_arrive(struct task_struct *t)
	157	{
	158	unsigned long long now = sched_clock(), delta = 0;
	159
	160	if (t->sched_info.last_queued)
	161	delta = now - t->sched_info.last_queued;
	162	sched_info_dequeued(t);
	163	t->sched_info.run_delay += delta;
	164	t->sched_info.last_arrival = now;
2d72376b	165	t->sched_info.pcount++;
425e0968 IM	166
	167	rq_sched_info_arrive(task_rq(t), delta);
	168	}
	169
	170	/*
	171	* Called when a process is queued into either the active or expired
	172	* array. The time is noted and later used to determine how long we
	173	* had to wait for us to reach the cpu. Since the expired queue will
	174	* become the active queue after active queue is empty, without dequeuing
	175	* and requeuing any tasks, we are interested in queuing to either. It
	176	* is unusual but not impossible for tasks to be dequeued and immediately
	177	* requeued in the same or another array: this can happen in sched_yield(),
	178	* set_user_nice(), and even load_balance() as it moves tasks from runqueue
	179	* to runqueue.
	180	*
	181	* This function is only called from enqueue_task(), but also only updates
	182	* the timestamp if it is already not set. It's assumed that
	183	* sched_info_dequeued() will clear that stamp when appropriate.
	184	*/
	185	static inline void sched_info_queued(struct task_struct *t)
	186	{
	187	if (unlikely(sched_info_on()))
	188	if (!t->sched_info.last_queued)
	189	t->sched_info.last_queued = sched_clock();
	190	}
	191
	192	/*
	193	* Called when a process ceases being the active-running process, either
	194	* voluntarily or involuntarily. Now we can calculate how long we ran.
	195	*/
	196	static inline void sched_info_depart(struct task_struct *t)
	197	{
	198	unsigned long long delta = sched_clock() - t->sched_info.last_arrival;
	199
	200	t->sched_info.cpu_time += delta;
	201	rq_sched_info_depart(task_rq(t), delta);
	202	}
	203
	204	/*
	205	* Called when tasks are switched involuntarily due, typically, to expiring
	206	* their time slice. (This may also be called when switching to or from
	207	* the idle task.) We are only called when prev != next.
	208	*/
	209	static inline void
	210	__sched_info_switch(struct task_struct prev, struct task_struct next)
	211	{
	212	struct rq *rq = task_rq(prev);
	213
	214	/*
	215	* prev now departs the cpu. It's not interesting to record
	216	* stats about how efficient we were at scheduling the idle
	217	* process, however.
	218	*/
	219	if (prev != rq->idle)
	220	sched_info_depart(prev);
	221
	222	if (next != rq->idle)
	223	sched_info_arrive(next);
	224	}
	225	static inline void
	226	sched_info_switch(struct task_struct prev, struct task_struct next)
	227	{
	228	if (unlikely(sched_info_on()))
	229	__sched_info_switch(prev, next);
230	}
231	#else
232	#define sched_info_queued(t) do { } while (0)
233	#define sched_info_switch(t, next) do { } while (0)
75d4ef16	234	#endif /* CONFIG_SCHEDSTATS */
425e0968	235